Confirmation Bias

I'm as optimistic as the next guy. Even more reason to think about what can go wrong.

There is a tremendous amount of excitement about the Department of Government Efficiency. The positive vibes are contagious. As we have suggested previously, this isn’t just about government. The drive to weed out bureaucracy and to make things run more smoothly is spreading.

The notion that we have to live with sclerotic processes because, you know, institutional history? It is dead. I have argued elsewhere that it is one of the drivers for the rally in risk assets. Bubbles start with good ideas. What happens when multiple good ideas arrive in waves? People look at AI as the unique force propelling equity markets higher. What if the imperative to de-bureaucratize is compounding the first wave with one of its own? What if the demographic impulse of the Millennial/GenZ transition into meaningful work is inflecting to a steeper trajectory as they enter their most productive years? What if we start to see the US consumer start to save and invest more at the same time? If all of these positive developments coincide, they compound. The natural instinct to descry the bubble may be premature.

It is at times like these that we must look for the hole in the boat. We are most likely to see only signs that reinforce confidence in our core optimistic hypothesis. This is known as confirmation bias. To fight this tendency, we must seek actively the ways in which we could be wrong and then look for supporting evidence of these countervailing forces.

One way to do this is to use a decision-making technique called a pre-mortem. Before we do something, as a way to surface reasons why a project may fail, we write a detailed letter from the perspective of our future self, explaining its failure to our stakeholders. Here’s Farnam Street with an explanation.

“Before finalizing any major decision—whether launching a product, making a key hire, or entering a new market—gather your team for a structured imagination exercise. Split them into two groups. The first group time-travels one year into the future where the initiative has completely failed: the product flopped, the hire was a disaster, the market expansion burned millions. The second group envisions remarkable success. Each person writes a detailed story explaining exactly how and why their scenario unfolded.”

Let’s go through this thought experiment for the DOGE effort.

As the second Trump administration ends, we can say that the scale and complexity of the problem of government bureaucracy overwhelmed the DOGE team. It did not make the government more efficient. It did not produce significant cost savings. DOGE leadership was unprepared for the kind of opposition it faced from an entrenched, unionized civil service. If anything, we should have seen this coming based upon the failure of the first Trump administration to do anything more than briefly arrest the growth of the administrative state.

Specifically, when it came to rulemaking, we know that the first Trump administration had the worst record dating back to President Clinton. As the Wall Street Journal reported, “About 57% of challenged major rules under his administration lost in court, which is the highest percentage since Bill Clinton’s second term, according to analysis from the Institute for Policy Integrity at the New York University School of Law.” DOGE did not improve on this record in Trump’s second term.

The DOGE team did not understand the reasons for the previous disappointing legal track record. It wasn’t enough to identify the processes and regulations that were superfluous. That’s difficult enough. To do this with permanent effect, they needed to prepare for the ineluctable legal challenges. What’s more, they had to do so in an unfavorable legal context. Most of the agencies they sought to reform fell under the jurisdiction of the DC Circuit Court of Appeals, a panel dominated by Democrats. DOGE wasn’t going to be able to appeal everything to the Supreme Court. The Supreme Court’s docket is limited.

There were civil servants who could have helped them. These men and women knew how to navigate the process of surviving a legal challenge such as developing the information a winning strategy would require. However, many of them were eliminated in waves of rolling job cuts. Musk and Ramaswamy set quantitative targets. Consultants and well-intentioned DOGE volunteers identified individuals to lay off. Unfortunately, as in many private sector mass lay-offs, they fired competent, sympathetic people who could have helped them advance their cause in addition to letting go some of the dead wood. Survivors became intense opponents of the de-bureaucratization initiative as a consequence of the layoffs.

DOGE underestimated the political benefit that some of their opponents sought to obtain from their failure. Given the hype around Musk and Ramaswamy before Trump took office a second time, the implication was clear. If DOGE failed, then the task would be deemed impossible. The dominance of government over the private sector and every aspect of American life would be complete.

Fortunately for their opponents, Musk and Ramaswamy had one playbook: making ambitious promises and promoting their ability to deliver. This had worked for them in the private sector. Even when Musk failed to hit his lofty entrepreneurial targets, he still succeeded in growing his financial capital. Politics is a different game. It is a game of gotcha. DOGE’s political capital evaporated when its promises failed to deliver on what proved to be an impossible timeline. Their political opponents helped accelerate this decline.

The volatility of their progress spilled over into the vibe, introducing volatility into the financial markets. When initial reports of their disappointing progress started to break from an antagonistic mainstream media, this produced large downdrafts in equity prices, in turn taken to be a “pricing-in” of their ultimate failure. Then, they would respond with some positive press release or court victory or cost savings and prices would rebound. Investors with weak hands lost money.

Musk and Ramaswamy exaggerated the impact of Loper Bright. This was the case that overturned the Chevron doctrine. Under Chevron, the previous norm for courts was to accept automatically the recommendation of agency experts on legal questions related to the administrative state. After Loper Bright, courts could pay attention to the expert opinions of the civil service, but judges would decide the merits of the case based upon the law as it was written. This pivotal decision and others such as Jarkesey that contested the power of the agencies were misinterpreted by many as an opportunity for the Executive Branch to take control of agencies that thought of themselves as independent. In practice, through a series of court decisions, it became clear that the practical impact of Loper Bright was to restore the power of Congress to write legislation that dictated the rules. This meant that the sustained success of DOGE’s efforts required Congress to pass significant and detailed legislation. Unfortunately, the GOP had insufficient Congressional control to make this happen. The early prospects of bipartisan consensus on increased administrative efficiency dissolved when the Democratic Party leadership realized that a failure of DOGE would inure to their significant political benefit and that such a demise was not only possible but likely. The GOP loss of control of the House and the narrowing of their Senate majority in the mid-term elections of 2026 eliminated the already dimming prospect of legislative support for administrative reform.

Many of the regulations and processes they sought to strip were a product of the risk aversion of the public service. In an organization in which the individual civil servant has no personal upside other than moderate career advancement towards a padded retirement, she faces only downside when things go wrong. This, combined with years of advocacy-driven litigation, created an environment of institutionalized defensiveness and reactionary response. Every time there was a scandal or an event, a new rule or process was put into place to address the embarrassment. Agencies were complex tapestries of bureaucracy. Even when agencies reversed themselves, their targets kept complying with the now defunct rules and processes.

DOGE cuts had iatrogenic effects because there were unanticipated multi-dimensional connections between rules and processes that Musk and Ramaswamy could not fathom. The bureaucracy was opaque, but underneath its black surface, an ecosystem of tremendous complexity teemed with life. As DOGE started acting on what they could, they changed the dynamic of the ecosystem in unpredictable ways. Some of it worked well. Much of it made things worse, as the ecosystem reacted, protecting itself from the invasive species of the reformer in ways that DOGE staffers did not anticipate. Making things worse, DOGE stimulated the already high levels of risk aversion in the civil service to new hysterical levels.

One irony of the DOGE effort is that their installation of Generative AI agents into each agency’s system started to identify the complexity and make recommendations for effective legislation, just as the political capital of the reform movement exhausted itself heading into the 2028 Presidential election cycle.

In short, DOGE may have worked decades earlier. The government had reached the point of no bureaucratic return well in advance of their arrival.

The purpose of the pre-mortem is two-fold.

First, it is part of a necessary debate to decide whether to proceed with the project. In this case, the decision has been made. DOGE is happening.

Second, in identifying ways in which things can go wrong, it enables project leadership to adjust the plan to minimize specific dimensions of risk.

We won’t know if DOGE leadership is doing this now. They may have plans about who to cut and who to retain. They may have obtained bipartisan legislative backing. They may have a sophisticated litigation strategy, one that has evolved from the poor track record of the first Trump administration. Perhaps, they have attracted the right people to execute.

I hope they succeed. I want to believe that DOGE is prepared. Even if they are not, I think that they will pivot quicky based on data. That is what entrepreneurs do.

Our best hope is that Musk and Ramaswamy cannot tolerate the reputational damage failure would bring.

I’ll take that bet.